lbhdc

No, I was a member of, and worked with a number of unions in heavy industry. After that experience I would never join or work with another union.

No, I don't talk to it like it is a person.

Can you share more about your setup?

I have a simple setup on GCP. I am using Cloud Build with the companion Github app to trigger builds on branch updates.

I like it because I am deploying to GCP, and storing containers in Artifact Registry. Cloud Build has good interop with those other products and terraform, so its pretty convenient to live with.

The pipelines themselves are pretty straight forward. Each step gets an image that it is executed in, and you can do anything you want in that step. There is some state sharing between steps, so if you build something in one step, you can use it in another.

I am pulling from a few registries, but trying to move everything to a private registry.

In podman, you have to use the "full path" to work with docker hub. Eg `docker.io/library/nginx`.

Oh hey! I have used your activity pub library, it's very nice :)

Its been a while, so all my telemetry has since expired, but there was no meaningful difference in time.

I was prepared to roll it all back, but I never ended up running into problems with it. It's just something that happens in the background that I don't have to think about.

Last year I transitioned all of my personal projects to use podman. The biggest surface area was converting CI to use podman to build my docker files, but also changed out tooling to use it (like having kind use it instead of docker).

For the most part this worked without issue. The only snag I ran into was my CI provider can't use oci formatted images. Podman lets you select the format of image to build, so I was able to work around this using the `--format=docker` flag.

Good point, it definitely isn't hands off from what I have seen. Just fewer people needed to manage those integrations.

I find this take interesting in the present. I work at $BIGCO that has done lots of ML in house for many years. I have noticed more and more ML projects being killed because their work is superseded by external model vendors. Why spend millions on a team to build a new model, when you can pay thousands to rent someone elses?

I Am a Strange Loop, by Hofstadter. I thought his idea of consciousness as a gradient was really interesting. Hofstadter said this idea made him a vegan. It did the opposite for me, eating a person has more or less the same moral perils as eating a cow. It didn't change my mind about eating animals, but eating people seems more or less okay.

I think what this misses is that TikTok will lose the ability to pay content creators. It may also cause legal repercussions for advertisers who want to spend on the platform.

From TikToks perspective, they may want the hard decouple so their users pressure their representatives. If they make a pwa or some other means to subvert the ban available it could undermine peoples desire to act.

I think it's one of those topics where most people feel too strongly about it to discuss rationally and be open to having their minds swayed.

It seems more like a sign of the current hiring climate. There have been a lot of layoffs and contractions in tech, and so companies are getting flooded with applicants (many of whom may not be accurately representing their skills).

My assumption is that these tests are lower effort ways to effectively weed out a large portion of applicants.

I think the popularity of Wordpress comes from the access it gives to non-technical people (plus ecosystem). I am sure someone is smart enough to write a CMS that can effectively do that under the hood, but I don't think it matters to the people who choose Wordpress.

> Streaming is too cheap

People aren't willing to pay more for music. The music industry learned that throughout the 2000s, and why things ended up the way they are. If the price increases, the path of least resistance becomes piracy.

We use backstage (backstage.io) to manage our infra. It has plugins that track costs and attribute them to individuals and teams. That gets aggregated and is used to forcast costs for projects/teams/whatever.

You can do click-ops in the UI (which then generates yaml in a repo), or you can write special yaml files in your repo yourself. These yaml files define the owner (team entity) or the individual that the cost originates from. Its a mostly automated process.

Since each resource an application uses is known, anomalies can be tracked down and attributed. So, for example, if someone starts serving big files from anywhere other than the CDN and blowing up egress costs, the source and root cause are easy to identify.

Backstage has a "lifecycle" tag for the resources you spin up (experimental is the default). If you spin stuff up that isn't tagged as being in production they get auto deleted after a period of time (you get a bunch of emails about it beforehand). That cleans up experiements or other test infra that people have forgotten about.

You are right, that is still and ad. Platform ads are really common for anyone running their own ad exchange. Sometimes its just to use up supply, other times its to meet in house goals.

I would totally pay for this if there were a real advantage (fewer applicants, guaranteed face to face screening, etc).

Has this been trading real money? How much money has it made?

How are you handling slippage?

No, I stopped even reading my email unless something prompts me to retrieve something from it.

They have a job listing looking for engineers. I have no idea if this role would work on HN or if the tech stack is relevant to HN.

It lists the tech stack as Rails, React, Postgres.

https://www.ycombinator.com/careers?ashby_jid=ef00c8d1-76e7-...

They have a paywall now, you can still view their free articles without an account.

My understanding is that most publications like this have declining financials. More competition, functional replacements in the market, declining search traffic, changing audience preferences. The pay wall is to try to re-capture some of that lost rev to keep things going. They have some free articles about this and have talked about it on their podcast.

Demand for machine learning engineers at most companies drops by 2026. I speculate that the demand for in-house models drops as the gulf between them and the major players grows. I think it will be (is?) easier and cheaper to rent time on someone else's model.

The ancillary tasks those MLEs were doing (data prep, model perf analysis, etc) will fall on other nearby roles (backend/data eng or data annalists). Ultimately I speculate that all of this work will end up being another integration backend/generalist engineers will handle.

That has been my experience with zig too, it sparks a lot of joy.

That is an interesting question. I am curious to know too.

If I were guessing, I would think their conference would skew towards tech professionals. Their articles show up on news aggregator apps, like google news, that probably brings in a ton of non-tech or tech adjacent people.

I think its really hard to cultivate a good comment section. Its pretty low effort to make off hand emotive comments ("this is bad and you should feel bad" kind of comments). Negative engagement is the most engaging engagement, so any reaction to that can encourage that behavior.

I think through social norms that get reinforced with moderation helps discourage some of that. That leaves room for people to ask questions and create discussion about the ideas in the article.

This is an example of a comment thread on the verge that I thought was interesting. If you view the full discussion you will find a bunch of uninteresting emotive comments as well.

https://www.theverge.com/2024/12/10/24318259/gm-cruise-shutd...

That is sick, I wanna build out something of a similar size (sans gpus). How much does colocation cost in your area?

I will start by saying that monetizing writing is in a difficult place right now. I appreciate that making this work is hard, and there is no silver bullet.

I assume I am in their target market, I am an SWE that likes consuming industry news. When I land on their webpage though, I struggle seeing information worth staying for.

The current top story is about Googles quantum chip that was on the front of HN all day yesterday. Their article is a summary of the release yesterday with what seems to be a summary of some arguments about multiverses, and a brief description of qubits vs bits.

It feels stale, I read about this yesterday from the primary source.

The article itself feels very surface level with some sensational dressing that feels like it is for clicks. It makes me wonder who this is written for? It seems too detailed for non-technical people (qubits vs bits), but doesn't go deep enough for technical readers (multiverses are deff real, trust me!).

That is just a single article, but when I look around, it doesn't feel like an outlier to me.

The value of actually visiting their website feels low. I think part of the problem is I don't read the content as "authentic". By that, I think their delivery leans too far to the click baity side. I recognize everyone is playing that game, I just don't think they are striking the right balance.

After I finish reading an article, it's nice to read thoughtful discussion about it. Its what makes HN so good. Slapping a comment section on it, isn't really enough though. Comments that are just emotive at the subject, or simply non-sequiters are no fun to read and having to wade through that lowers the value of the experience. The verge has a website with comments that _can_ be worth reading because some people will actually talk about the content.

Can you describe your GPU setup? I have been super interested to get some coloc space, but have a ton of questions.

Would a performance impact be persuasive?

Ultimately they are trying to use that data to answer "when we do x the business makes y dollars". The data they get is pretty granular, and shows the users behavior across multiple platforms making it very hard/impossible to replicate inhouse.

So, the trade off seems to be performance (privacy, lower bounce rate, possibly better ranking) vs insight into how your product is being used and what some of the other things your users are interested in.

Making a convincing argument to go without for some perf is going to be a really hard sell. Especially if those people have new laptops and fast internet. I think if you are going to go this route you need to yoke that idea to the money. "Saving x resources lowers our cloud bill by y%" or "Our product would be a good fit for people in this country, but they tend to have slow mobile internet, increasing our performance will unlock that market".

One compromise could be using 'aggregator pixels'. Effectively you add a single tracking script to your site, and they forward that data to all the different analytics companies your colleagues want to use. Metarouter is one that comes to mind (no affiliation).